Algorithm Baseline for L1 Product and Calibration

Transcription

1 Algorithm Baseline for L1 Product and Calibration Oliver Reitebuch Uwe Marksteiner, Karsten Schmidt Dorit Huber, Ines Nikolaus Alain Dabas, Pauline Martinet in close cooperation with ESA, Airbus Defense & Space, L2 Algorithm Team from ECMWF, KNMI

2 Outline of the talk How are winds measured by ALADIN and can be retrieved? ESA/ATG-Medialab Why is a calibration needed for ALADIN and how is it performed? What are the processing steps and challenges for wind retrieval up to Level 1? 2

3 It started here at ESRIN, Frascati in 1968 with Giorgio Fiocco ( ) at European Space Research Institute, Frascati Benedetti-Michelangeli et al. (2003). Annals Geophysics 3

4 Aeolus the first wind lidar in space - first time that retrieval algorithms for spaceborne wind lidars are developed High requirement on random error (precision): 1 m/s (0-2 km) to 2 m/s (2-16 km) HLOS => mainly determined by photon Poisson noise Algorithms for quality control and error quantifiers are affected Very demanding requirements for the systematic error (accuracy): bias <0.4 m/s and linearity error <0.7 % of actual wind speed => mainly determined by instrument stability and calibration H max : km ΔH: m Strong influence of algorithms for wind retrieval, calibration, ground-correction and bias characterisation and correction Wind products need to be available for NWP users within 3 hours after observation => different to other lidar missions and science-driven missions Fig.: adapted from ESA (2008) Monitoring of Aeolus wind products with NWP models, but strong need to provide consolidated and validated wind products within some months after launch 4

5 Principle of Rayleigh and Mie spectrometer for ALADIN Doppler -Equation: f 2 f 1 m/s 5.64 MHz 2.37 fm nm THz 0 v LOS c 1 pm 422 m/s Rayleigh wind sensitivity 0.3% / m/s => Determine signal intensity with an accuracy in the order of 0.05% to 0.5% => offset corrections and calibration Mie wind sensitivity 18 m/s per pixel + Mie fringe width 30 m/s Fig.: Reitebuch (2012): Wind Lidar for Atmospheric Research, in Springer Series => Determine centroid of a signal, which is 30 m/s broad (FWHM) with an accuracy of 1/100 to 1/20 of a pixel width => high SNR and QC 5

6 16 Pixel 16 Pixel Why are calibrations needed for direct-detection wind lidar? Rayleigh spots Measure R for v LOS =0 m/s (nadir pointing) and vary Δf by changing laser frequency I A I B 16 Pixel Mie fringe R I I A A I I B B or R I I Instrument Response R A B Measure R(Δf) Calibration R(Δf) Doppler frequency f and LOSwind vlos v f 2 f LOS 0 c R : xˆ 0 centroidpixel Compute R(Δf) x 0 16 Pixel Lineshape model I(f) Rayleigh-Brillouin I(f)=f(T,p,h) and laser lineshape I(f) Instrument function T(f) for Rayleigh and Mie spectrometer 6

7 intensity [a.u.] Rayleigh and Mie Instrument functions are determined on ground and in-orbit A B 844, , , , , ,765 wavemeter absolute frequency [THz] FSR 10.9 GHz f c filter crosspoint 1 GHz calibration range A2D measurements and fits with Airy- Model: Airy_fit_mit_defekten_alles_ y No weighting Chi^2/DoF = R^2 = function convolved with Gaussian Defect Function + A ± Q ± Rb ± FSRb ± f0b ± defectb ± R ± f ± FSR ± defect 7340 ± Fizeau reflection STD from several characterisations: FSR: 23 MHz (0.2%) FWHM: 30 MHz (1.8%)

8 Calibration of Rayleigh/Mie spectrometer response Characterization of Internal and Atmospheric Response for different laser frequencies for Mie and Rayleigh spectrometer => fit coefficients used for L1 wind retrieval 8

9 Chain of processors for Aeolus ALADIN instrument modes in E2S and L1B 9

10 The Aeolus End-to-End Simulator E2S Orbit Atmosphere ALADIN Data + Pointing + Ground instrument formatter ADAM project: Fig. A.G. Straume, Th. Kanitz (ESA) 10

11 ALADIN Mie+ Rayleigh Main processing steps for wind retrieval up to Level 1 f f f f ALADIN LOS, atmos LOS, sat offset f atmos f internal platform Doppler Lidar AOCS: Attitude+Orbit Control System DEM: Digital Elevation Model L1 Wind Mode Data all instrument related corrections and calibrations both Mie and Rayleigh winds no atmospheric corrections, e.g. temperature, pressure, aerosol cross-talk to Rayleigh => L2 product no scene classification or grouping => L2 product Additional L1 Data Geolocation Signal amplitudes and SNR Error quantifiers and quality flags Geolocation, pointing, instrument data Ground-return signal and velocity 11

12 Unique calibration target for space-/airborne wind lidars: the (non-)moving ground A2D in Greenland 2009 Clouds or Ground? 28 km ALADIN range gates m V LOS,SAT ALADIN ground speed v w v (1 w ) v ground ground LOS, Sat ground atmos, cloud w f ( R, R,, ) ground atmos, cloud land V cloud V atmosphere land: albedo ρ 0.8 (ice) 0.03 (soil) sea ρ(v, θ) correction of residual satellite speed Mie and Rayleigh response calibration in nadir pointing mode ALADIN bias corrections V land = 0 m/s V sea 0 m/s

13 Key challenges for wind retrieval from Aeolus processing purpose challenge Rayleigh offset corrections wind derived from signal I 1 % I = 3.5 m/s LOS => high SNR + systematic errors in I <0.05% Mie pixel corrections (=spectral corrections) Mie and Rayleigh response cal. with atmos. and ground returns Ground detection and quality-control of ground velocity LOS wind derived from v LOS = (f a -f i )*λ 0 /2 fringe-imaging for Mie; wind derived from centroid input to L1 and L2 wind retrieval Mie/Rayleigh response cal., v SAT correction, and bias characterisation principle of wind retrieval for L1 and L2 1 pixel = 18 m/s LOS => centroid accuracy 0.05 pixel => high SNR random error during calibration => systematic error for wind retrieval; characterisation of difference in optical path internal / atmosphere coarse range gates => atmospheric contribution no monitoring of absolute laser wavelength, no locking to molecular line, no differential wind retrieval v LOS,c = v LOS -v 0 Bias corrections for L1 and L2 winds characterisation of the bias correction coefficients 13

14 Summary and Conclusion 2 different approaches for wind retrieval in ALADIN for molecular Rayleigh return and aerosol/cloud return ESA/ATG-Medialab Calibration from instrument response to frequency is needed and 2 approaches are currently implemented Aeolus is the first wind lidar and it will be the first time that retrieval algorithms will be challenged by real atmospheric observations from space Reitebuch, Huber, Nikolaus: Algorithm Theoretical Baseline Document ATBD, V 4.1, July